Dictation and transcription
Dictation lets you speak a prompt instead of typing it. The browser records your voice, POM AI transcribes it, and the text is added to the prompt field you were working in.
It is an input method for prompts, not a content tool. Nothing is published, and the transcript never reaches a post by itself.
Where it appears
The Dictate prompt button sits above the prompt field on the tools that accept a written prompt, including:
- Posts → Write posts with AI
- Pages → AI texts generator
- Media → AI generate images
- Media → AI improve image
- Media → AI remove image background
There is nothing to enable separately. If a tool is enabled and has a prompt field, its dictation control is there.
Guides
- Dictate text — recording a prompt.
- Allow microphone access — the browser permission it needs.
- Review and insert dictated text — what happens to the transcript.
- Transcribe an audio recording — limits, formats and what the service returns.
- Dictation language and punctuation — what to expect from the text.
- Audio privacy — what leaves your site.
- Troubleshoot dictation.
How it works, briefly
- You press Dictate prompt, or use Ctrl+Shift+.
- The browser asks for microphone permission the first time.
- Recording starts, with a live status line counting elapsed time against the limit.
- You stop, or the limit stops it for you.
- The recording is sent to your site, then on to the POM AI account service.
- The transcript is appended to the prompt field.
The limits worth knowing
| Limit | Value |
|---|---|
| Maximum recording length | 3 minutes |
| Maximum recording size | 4 MB |
| Credit cost | 6 credits per started minute |
Because the cost is per started minute, a ten-second note and a fifty-second one both cost six credits. Dictating one longer prompt is cheaper than dictating three short ones.
Unlike the generation tools, dictation does not show an estimate or a confirmation dialog. It transcribes as soon as you stop recording. The cost is small and bounded — at most eighteen credits for a full three-minute recording — but it is charged without asking.
What it needs
- A browser that supports audio recording, and a granted microphone permission.
- A valid POM AI license, since transcription is an account service.
- Credits in the account.
- The
edit_postsorupload_filescapability, matching the tool you are using.
What it does not do
- It does not transcribe uploaded audio files. There is no upload field; the recording is made in the browser.
- It does not transcribe video, podcasts or meeting recordings.
- It does not write into a post, only into a prompt field.
- It does not translate. The transcript comes back in the language you spoke.
- It does not punctuate perfectly, or format. See Dictation language and punctuation.
When dictation is worth using
It suits describing something long or exploratory — the kind of prompt where you would ramble in a briefing and then tidy it up. Speaking a paragraph is faster than typing it, and the tidying happens in the prompt field before you generate.
It is not worth it for a short prompt you could type in ten seconds, because the minimum charge is the same as for a full minute.